Connections¶

Connections document the infrastructure-level network paths between servers and entities. While Interfaces describe logical data flows between applications, Connections describe the physical network routes: which servers communicate, over which protocols and ports.

Getting started¶

Navigate to IT Landscape > Connections to see your connection registry. To create a connection:

Click Add connection. The connection page opens with an empty title in the header. The rest of the page is dimmed.
Type a descriptive name.
Press Enter, or click outside the title. Kadratis creates the connection at once, server to server, without protocols or endpoints. Its reference (CONN-12) shows and all tabs open.

Each field then saves on its own. A connection needs: - Topology: Server to server or Multi-server - Endpoints: A source and a destination for Server to server (each a server, a cluster or an entity), or at least two servers for Multi-server - Protocols: At least one network protocol

Press Escape on the empty title to leave without creating anything.

Tip: Connections can be linked to interface bindings to show which infrastructure supports each application integration.

Creating or cancelling¶

While the connection is being created, its header shows Create and Cancel. They stay until you leave the connection.

Create checks that the connection has at least one protocol and both endpoints, or two servers for a multi-server connection. A message names what is missing, and the first missing field gets the focus. When nothing is missing, Kadratis keeps the connection and closes it. You go back to where you started
Cancel deletes the connection. If you entered more than the name, a confirmation lists what will be deleted. You can cancel your own creation without the right to delete connections

Leaving the connection another way while something is missing asks what to do. Complete it keeps you on the connection, on the first missing field. Delete it deletes the connection.

Working with the list¶

The list gives you a filterable overview of every connection in your registry.

Default columns: - Reference: The connection reference, such as CONN-12 (click to open workspace) - Name: Connection name (click to open workspace) - Topology: Server to server or Multi-server - Source / Destination: The connected endpoints - Protocols: Network protocols shown as chips - Criticality: Business importance, which may be derived from linked interfaces - Data class: Data sensitivity level - PII: Whether personal data traverses this connection - Risk: Manual or Derived (shows the number of linked interfaces) - Lifecycle: Current status - Created on: When the record was created

Additional columns (via column chooser): - Servers: Count of servers in a multi-server connection

Filtering: - Quick search: Searches across connection fields - Column filters: Topology, Lifecycle, Criticality, Data class and PII use checkbox lists, so you can pick several values at once. Criticality and Data class also offer (Blank) for connections without a value - Criticality, Data class and PII filter and sort on the value shown. For a derived connection, that is the value inherited from its interfaces

Sorting: - Criticality and Data class sort by severity, in the order of the levels in Classifications and continuity. Connections without a value come last

Actions: - Add connection: Create a new connection (requires infrastructure:member) - Delete connection: Remove selected connections (requires infrastructure:admin). The hops of each deleted connection are deleted with it. Their servers, interfaces and documents stay; only their links to the connection go. To delete a single connection, open it and use Delete in its header, which needs infrastructure:member (see The Connections workspace).

Connection types¶

Server to server¶

A direct connection between two endpoints. Each side is a server, a cluster or an entity:

Source: Where traffic originates
Destination: Where traffic terminates
A side holds one server, cluster or entity
The source and the destination cannot be the same asset

Multi-server¶

A connection between several servers, such as a replication between nodes or a mesh. Select at least two servers or clusters.

Connections on a cluster¶

A source, a destination, a server of a multi-server connection or the equipment of a hop can be a cluster. The connection then applies to the members in service of the cluster. A member added later gets it too. A member taken out of the cluster loses it. A retired member gets nothing, and a retired cluster passes nothing on.

The pickers mark a cluster with an icon and list its members under its name.

When you pick a server that belongs to a cluster, Kadratis asks Connect to the cluster?. Connect to the cluster puts the cluster on the connection, for all its members. This server only keeps the server. Kadratis does not ask when the other side is already that cluster or one of its members: the connection stays on the server.

A multi-server list that holds a cluster and one of its members keeps the cluster only. A message names the members removed from the list. A list must keep at least two servers.

The Network tab of each member lists the connections of its clusters, and the Connection Map draws the members of a cluster.

The Connections workspace¶

Click any row to open the workspace. It has three tabs: Overview, Path and History (the changes made to the connection, see Record history). The properties panel on the right holds Lifecycle, Topology, Risk mode, Criticality, Data class, Contains PII and Created on. Edits save automatically.

Delete, a trash icon in the header, deletes the connection after a confirmation (infrastructure:member). The confirmation counts the hops deleted with the connection and the interface bindings it is unlinked from. The interfaces themselves stay.

Overview¶

Description: Why this connection exists and what it carries
Endpoints: The source and the destination, or the servers of a multi-server connection. See Connection types
Protocols: One or more network protocols, from your Connection Types settings, with their typical ports
Risk derivation: In derived risk mode, the interfaces the risk comes from, with their criticality, data class and PII
Linked interfaces: The interface bindings linked to this connection. Link existing links another one

Path¶

The Path tab documents the network path of the connection, hop by hop: reverse proxies, firewalls, load balancers.

Each hop has:

A function, such as Firewall, Reverse proxy or Load balancer
Equipment: A server, a cluster or an entity, or nothing when the equipment is not documented. A server that belongs to a cluster raises the same question as an endpoint. See Connections on a cluster
Protocols (override) and Port: When they differ from the connection
Notes

Add hop adds a hop at the end. The arrows reorder the hops and the bin removes one. Each change saves at once.

Risk¶

Risk mode, in the properties panel:

Manual: You set criticality, data class and PII directly
Derived: The values come from the linked interface bindings. The fields are read-only and show the highest values across the linked interfaces

Criticality uses the levels of Classifications and continuity. The default levels are Critical, High, Moderate and Low. Data class comes from your data classification settings.

Tips¶

Start with critical paths: Document connections for your most important applications first, then work outward.
Use Derived risk mode: Let the system calculate criticality from the interfaces that use each connection. It saves effort and stays current as interfaces change.
Link to interfaces: Connecting your infrastructure to interface bindings gives you end-to-end traceability from application data flows down to network routes.
Document protocols accurately: Good protocol data makes firewall rule reviews and security audits significantly easier.